Lecture Overview. Lecture 3 Design Philosophy & Applications. Internet Architecture. Priorities

Transcription

1 Lecture Overview Lecture 3 Design Philosophy & Applications David Andersen School of Computer Science Carnegie Mellon University Networking, Spring ! Last time:» Protocol stacks and layering» OSI and TCP/IP models» Application requirements from transport protocols! Internet Architecture! Application examples.» ftp» http! Application requirements.» ilities» Sharing 2 Internet Architecture Priorities! Background» The Design Philosophy of the DARPA Internet Protocols (David Clark, 1988).! Fundamental goal: Effective network interconnection! Goals, in order of priority: 1. Continue despite loss of networks or gateways 2. Support multiple types of communication service 3. Accommodate a variety of networks 4. Permit distributed management of Internet resources 5. Cost effective 6. Host attachment should be easy 7. Resource accountability! The effects of the order of items in that list are still felt today» E.g., resource accounting is a hard, current research topic! Let s look at them in detail 3 4

2 Survivability Fate Sharing! If network disrupted and reconfigured» Communicating entities should not care!» No higher-level state reconfiguration» Ergo, transport interface only knows working and not working. Not working == complete partition.! How to achieve such reliability?» Where can communication state be stored? Failure handing Net Engineering Switches Network Replication Tough Maintain state Host Fate sharing Simple Stateless Connection State No State State! Lose state information for an entity if (and only if?) the entity itself is lost.! Examples:» OK to lose TCP state if one endpoint crashes NOT okay to lose if an intermediate router reboots» Is this still true in today s network? NATs and firewalls! Survivability compromise: Heterogenous network -> less information available to end hosts and Internet level recovery mechanisms Host trust Less More 5 6 Types of Service Varieties of Networks! Recall from last time TCP vs. UDP» Elastic apps that need reliability: remote login or » Inelastic, loss-tolerant apps: real-time voice or video» Others in between, or with stronger requirements» Biggest cause of delay variation: reliable delivery Today s net: ~100ms RTT Reliable delivery can add seconds.! Original Internet model: TCP/IP one layer» First app was remote login» But then came debugging, voice, etc.» These differences caused the layer split, added UDP! No QoS support assumed from below» In fact, some underlying nets only supported reliable delivery Made Internet datagram service less useful!» Hard to implement without network support» QoS is an ongoing debate 7! Discussed a lot of this last time -» Interconnect the ARPANET, X.25 networks, LANs, satellite networks, packet networks, serial links! Mininum set of assumptions for underlying net» Minimum packet size» Reasonable delivery odds, but not 100%» Some form of addressing unless point to point! Important non-assumptions:» Perfect reliability» Broadcast, multicast» Priority handling of traffic» Internal knowledge of delays, speeds, failures, etc.! Much engineering then only has to be done once 8

3 The Other goals Accountability! Management» Today s Internet is decentralized - BGP» Very coarse tools. Still in the assembly language stage! Cost effectiveness» Economies of scale won out» Internet cheaper than most dedicated networks» Packet overhead less important by the year! Attaching a host» Not awful; DHCP and related autoconfiguration technologies helping. A ways to go, but the path is there! But! Huge problem.! Accounting» Billing? (mostly flat-rate. But phones are moving that way too - people like it!)» Inter-provider payments Hornet s nest. Complicated. Political. Hard.! Accountability and security» Huge problem.» Worms, viruses, etc. Partly a host problem. But hosts very trusted.» Authentication Purely optional. Many philosophical issues of privacy vs. security.! Questions before we move on to the project? 9 10 FTP: The File Transfer Protocol Ftp: Separate Control, Data Connections user at host FTP user interface FTP client local file system file transfer FTP server! Transfer file to/from remote host! Client/server model» Client: side that initiates transfer (either to/from remote)» Server: remote host! ftp: RFC 959 remote file system! Ftp client contacts ftp server at port 21, specifying TCP as transport protocol! Two parallel TCP connections opened:» Control: exchange commands, responses between client, server. out of band control» Data: file data to/from server! Ftp server maintains state : current directory, earlier authentication FTP client TCP control connection port 21 TCP data connection port 20 FTP server! ftp server: port

4 Ftp Commands, Responses HTTP Basics Sample Commands:! sent as ASCII text over control channel! USER username! PASS password! LIST return list of files in current directory! RETR filename retrieves (gets) file! STOR filename stores (puts) file onto remote host Sample Return Codes! status code and phrase! 331 Username OK, password required! 125 data connection already open; transfer starting! 425 Can t open data connection! 452 Error writing file! HTTP layered over bidirectional byte stream» Almost always TCP! Interaction» Client sends request to server, followed by response from server to client» Requests/responses are encoded in text! Stateless» Server maintains no information about past client requests How to Mark End of Message? HTTP Request! Size of message! Content-Length» Must know size of transfer in advance! Delimiter! MIME style Content-Type» Server must escape delimiter in content! Close connection» Only server can do this 15 16

5 HTTP Request HTTP Request! Request line» Method GET return URI HEAD return headers only of GET response POST send data to the server (forms, etc.)» URI E.g. with a proxy E.g. /index.html if no proxy» HTTP version! Request headers» Authorization authentication info» Acceptable document types/encodings» From user » If-Modified-Since» Referrer what caused this page to be requested» User-Agent client software! Blank-line! Body HTTP Request Example HTTP Response GET / HTTP/1.1 Accept: */* Accept-Language: en-us Accept-Encoding: gzip, deflate User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0) Host: Connection: Keep-Alive! Status-line» HTTP version» 3 digit response code 1XX informational 2XX success! 200 OK 3XX redirection! 301 Moved Permanently! 303 Moved Temporarily! 304 Not Modified 4XX client error! 404 Not Found 5XX server error! 505 HTTP Version Not Supported» Reason phrase 19 20

6 HTTP Response HTTP Response Example! Headers» Location for redirection» Server server software» WWW-Authenticate request for authentication» Allow list of methods supported (get, head, etc)» Content-Encoding E.g x-gzip» Content-Length» Content-Type» Expires» Last-Modified! Blank-line! Body HTTP/ OK Date: Tue, 27 Mar :49:38 GMT Server: Apache/ (Unix) (Red-Hat/Linux) mod_ssl/2.7.1 OpenSSL/0.9.5a DAV/1.0.2 PHP/4.0.1pl2 mod_perl/1.24 Last-Modified: Mon, 29 Jan :54:18 GMT ETag: "7a11f-10ed-3a75ae4a" Accept-Ranges: bytes Content-Length: 4333 Keep-Alive: timeout=15, max=100 Connection: Keep-Alive Content-Type: text/html Cookies: Keeping state Cookies: Keeping State (Cont.) client Amazon server Many major Web sites use cookies Four components: 1) Cookie header line in the HTTP response message 2) Cookie header line in HTTP request message 3) Cookie file kept on user s host and managed by user s browser 4) Back-end database at Web site Example:» Susan accesses Internet always from same PC» She visits a specific e- commerce site for the first time» When initial HTTP requests arrives at site, site creates a unique ID and creates an entry in backend database for ID Cookie file ebay: 8734 Cookie file amazon: 1678 ebay: 8734 one week later: Cookie file amazon: 1678 ebay: 8734 usual http request msg usual http response + Set-cookie: 1678 usual http request msg cookie: 1678 usual http response msg usual http request msg cookie: 1678 usual http response msg server creates ID 1678 for user cookiespecific action cookiespecific action entry in backend database access access 23 24

7 Typical Workload (Web Pages) HTTP new features! Multiple (typically small) objects per page! File sizes» Why different than request sizes?» Also heavy-tailed Pareto distribution for tail Lognormal for body of distribution! Embedded references» Number of embedded objects = pareto p(x) = ak a x -(a+1)! Newer versions of HTTP add several new features (persistent connections, pipelined transfers) to speed things up.! Let s detour into some performance evaluation and then look at those features Packet Delay Packet Delay Store & Forward Cut-through Prop + xmit 2*(Prop + xmit) 2*prop + xmit! Sum of a number of different delay components.! Propagation delay on each link.» Proportional to the length of the link! Transmission delay on each link.» Proportional to the packet size and 1/link speed! Processing delay on each router.» Depends on the speed of the router! Queuing delay on each router.» Depends on the traffic load and queue size When does cut-through matter? Next: Routers have finite speed (processing delay) D B C B A A Routers may buffer packets (queueing delay) 27 28

8 A Word about Units Application-level Delay! What do Kilo and Mega mean?» Depends on context! Storage works in powers of two.» 1 Byte = 8 bits» 1 KByte = 1024 Bytes» 1 MByte = 1024 Kbytes! Networks work in decimal units.» Network hardware sends bits, not Bytes» 1 Kbps = 1000 bits per second» To avoid confusion, use 1 Kbit/second! Why? Historical: CS versus ECE. Delay of one packet Average sustained throughput Delay * + Size Throughput Units: seconds + bits/(bits/seconds) 29 * For minimum sized packet 30 Some Examples Sustained Throughput! How long does it take to send a 100 Kbit file?» Assume a perfect world» And a 10 Kbit file Throughput Latency 500 µsec 100 Kbit/s 1 Mbit/s Mbit/s ! When streaming packets, the network works like a pipeline.» All links forward different packets in parallel! Throughput is determined by the slowest stage.» Called the bottleneck link! Does not really matter why the link is slow.» Low link bandwidth» Many users sharing the link bandwidth 10 msec 100 msec

9 One more detail: TCP HTTP 0.9/1.0! TCP connections need to be set up» Three Way Handshake : Client Server SYN (Synchronize) SYN/ (Synchronize + Acknowledgement) Data 2: TCP transfers start slowly and then ramp up the! One request/response per TCP connection» Simple to implement! Disadvantages» Multiple connection setups! three-way handshake each time Several extra round trips added to transfer» Multiple slow starts bandwidth used (so they don t use too much) Single Transfer Example Performance Issues Client opens TCP connection Client sends HTTP request for HTML Client parses HTML Client opens TCP connection Client sends HTTP request for image Image begins to arrive 0 RTT 1 RTT 2 RTT 3 RTT 4 RTT Client SYN FIN SYN Server SYN FIN SYN Server reads from disk Server reads from disk! Short transfers are hard on TCP» Stuck in slow start» Loss recovery is poor when windows are small! Lots of extra connections» Increases server state/processing! Servers also hang on to connection state after the connection is closed» Why must server keep these?» Tends to be an order of magnitude greater than # of active connections, why? 35 36

10 Netscape Solution Persistent Connection Solution! Mosaic (original popular Web browser) fetched one object at a time!! Netscape uses multiple concurrent connections to improve response time» Different parts of Web page arrive independently» Can grab more of the network bandwidth than other users! Doesn t necessarily improve response time» TCP loss recovery ends up being timeout dominated because windows are small! Multiplex multiple transfers onto one TCP connection! How to identify requests/responses» Delimiter! Server must examine response for delimiter string» Content-length and delimiter! Must know size of transfer in advance» Block-based transmission! send in multiple length delimited blocks» Store-and-forward! wait for entire response and then use content-length» Solution! use existing methods and close connection otherwise Persistent Connection Solution Persistent HTTP Client sends HTTP request for HTML 0 RTT 1 RTT Client parses HTML Client sends HTTP request for image Image begins to arrive 2 RTT Client Server Server reads from disk Server reads from disk Nonpersistent HTTP issues:! Requires 2 RTTs per object! OS must work and allocate host resources for each TCP connection! But browsers often open parallel TCP connections to fetch referenced objects Persistent HTTP! Server leaves connection open after sending response! Subsequent HTTP messages between same client/server are sent over connection Persistent without pipelining:! Client issues new request only when previous response has been received! One RTT for each referenced object Persistent with pipelining:! Default in HTTP/1.1! Client sends requests as soon as it encounters a referenced object! As little as one RTT for all the referenced objects 39 40

11 Persistent Connection Performance Remaining Problems! Benefits greatest for small objects» Up to 2x improvement in response time! Server resource utilization reduced due to fewer connection establishments and fewer active connections! TCP behavior improved» Longer connections help adaptation to available bandwidth» Larger congestion window improves loss recovery! Serialized transmission» Much of the useful information in first few bytes May be better to get the 1st 1/4 of all images than one complete image (e.g., progressive JPEG)» Can packetize transfer over TCP Could use range requests! Application specific solution to transport protocol problems. :(» Solve the problem at the transport layer» Could fix TCP so it works well with multiple simultaneous connections More difficult to deploy Back to performance Bandwidth Sharing! We examined delay,! But what about throughput?! Important factors:» Link capacity» Other traffic! Bandwidth received on the bottleneck link determines end-to-end throughput.! Router before the bottleneck link decides how much bandwidth each user gets.» Users that try to send at a higher rate will see packet loss! User bandwidth can fluctuate quickly as flows are added or end, or as flows change their transmit rate. 100 BW Time 43 44

12 Fair Sharing of Bandwidth But It is Not that Simple! All else being equal, fair means that users get equal treatment.» Sounds fair! When things are not equal, we need a policy that determines who gets how much bandwidth.» Users who pay more get more bandwidth» Users with a higher rank get more bandwidth» Certain classes of applications get priority 100 BW Time Bottleneck Network Service Models Other Requirements! Set of services that the network provides.! Best effort service: network will do an honest effort to deliver the packets to the destination.» Usually works! Guaranteed services.» Network offers (mathematical) performance guarantees» Can apply to bandwidth, latency, packet loss,..! Preferential services.» Network gives preferential treatment to some packets» E.g. lower queuing delay! Quality of Service is closely related to the question of fairness.! Network reliability.» Network service must always be available! Security: privacy, DOS,..! Scalability.» Scale to large numbers of users, traffic flows,...! Manageability: monitoring, control,..! Requirement often applies not only to the core network but also to the servers.! Requirements imposed by users and network managers

13 Readings! End-to-end arguments in system design, Saltzer, Reed, and Clark, ACM Transactions on Computer Systems, November 1984.! The design philosophy of the DARPA Internet Protocols, Dave Clark, SIGCOMM